The relation between similarity and dissimilarity of meaning and similarity of context was analyzed for synonymous nouns. New semantic similarity and dissimilarity rating tests with an empirically determined series of linguistic anchors and conventional, arbitrarily anchored semantic similarity ratings were compared. Contextual similarity was elicited by a sorting test based on substitution and yielding d-primes. The study found reliable correlations between the d-primes and the different ratings for semantic similarity and dissimilarity of the synonymous nouns across a wide continuum of meaning. The data strongly supported a contextual hypothesis of meaning. The data endorsed the claim that people abstract a contextual representation from experiencing the multiple natural linguistic contexts of a word. Semantic similarity and dissimilarity rating formats with an empirically chosen series of linguistic anchors and a sorting test of contextual similarity yielded stronger support for a contextual hypothesis than did alternative methods of eliciting lexical and contextual similarity.